lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Nagendra Nagarajayya (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (SOLR-3816) Need a more granular nrt system that is close to a realtime system.
Date Mon, 10 Sep 2012 15:08:08 GMT

    [ https://issues.apache.org/jira/browse/SOLR-3816?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13452045#comment-13452045
] 

Nagendra Nagarajayya edited comment on SOLR-3816 at 9/11/12 2:07 AM:
---------------------------------------------------------------------

The attached patch provides realtime NRT that is close to a realtime system and offers fine
granular NRT at very high performance and offers a path to a true realtime system as 
and when the underlying IndexWriter system supports it. A performance of about 70,000 document
adds / sec has been seen on a user system with realtime queries (almost 1.5-2x improvement
over soft-commit)

The realtime nrt can be enabled by adding the following tag to SolrConfig.xml:

<realtime visible="milliseconds">true/false</realtime>

A true enables realtime nrt while false disables it.

The visible attribute controls the time a document may not visible in a search. Settng this
to 0 means new documents are visible immediately in searches. Very High performance can be

observed with visible around 150-200ms.
For eg: <realtime visible="150">true</realtime>

The other parameters supported are:

<deleteDuplicates>true/false</deleteDuplicates>

A true deletes duplicates while a false allows duplicates to exist until a commit. Setting
this to true may impact performance.

The same may be achieved by setting <maxBufferedDeleteTerms> to 1.


Note: 

1. The cache (Query Result Cache, etc.) needs to be disabled for realtime NRT. 
2. Increase the number of file descriptors to around 64k before starting Solr
	ie. ulimit -n 65536




                
      was (Author: nnagarajayya):
    The attached patch provides a realtime NRT that is closer to a realtime system and offers
a fine granular NRT, at very high performance and offers a path to a true realtime system
as 
and when the underlying IndexWriter system supports it. A performance of about 70,000 document
adds / sec has been seen on a user system with realtime queries (almost 1.5-2x improvement
over soft-commit)

The realtime nrt can be enabled by adding the following tag to SolrConfig.xml:

<realtime visible="milliseconds">true/false</realtime>

A true enables realtime nrt while false disables it.

The visible attribute controls the time a document may not visible in a search. Settng this
to 0 means new documents are visible immediately in searches. Very High performance can be

observed with visible around 150-200ms.
For eg: <realtime visible="150">true</realtime>

The other parameters supported are:

<deleteDuplicates>true/false</deleteDuplicates>

A true deletes duplicates while a false allows duplicates to exist until a commit. Setting
this to true may impact performance.

The same may be achieved by setting <maxBufferedDeleteTerms> to 1.


Note: 

1. The cache (Query Result Cache, etc.) needs to be disabled for realtime NRT. 
2. Increase the number of file descriptors to around 64k before starting Solr
	ie. ulimit -n 65536




                  
> Need a more granular nrt system that is close to a realtime system.
> -------------------------------------------------------------------
>
>                 Key: SOLR-3816
>                 URL: https://issues.apache.org/jira/browse/SOLR-3816
>             Project: Solr
>          Issue Type: Improvement
>          Components: clients - java, replication (java), search, SearchComponents - other,
SolrCloud, update
>    Affects Versions: 4.0
>            Reporter: Nagendra Nagarajayya
>              Labels: nrt, realtime, replication, search, solrcloud, update
>         Attachments: solr-3816-realtime_nrt.patch
>
>
> Need a more granular NRT system that is close to a realtime system. A realtime system
should be able to reflect changes to the index as and when docs are added/updated to the index.
soft-commit offers NRT and is more realtime friendly than hard commit but is limited by the
dependency on the SolrIndexSearcher being closed and reopened and offers a coarse granular
NRT. Closing and reopening of the SolrIndexSearcher may impact performance also.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org


Mime
View raw message